AWS: Increasing Engagement with Robot Arms

Situation

Amazon Web Services (AWS) operates Skills Centers that serve as a significant recruitment and talent pipeline for the company. These centers provide opportunities for the public to learn about various technologies, including robotics. To showcase the potential of robotics in the workplace, AWS required an interactive robot autonomy demonstration that could be easily operated by their Skills Center staff. The demo would serve as a powerful tool to promote the adoption of robotics and attract potential talent to AWS.

Problem

Designing and implementing a functional robot system is a challenging task, as it requires solving several complex problems. Pick-and-place demos, although seemingly simple, demand solutions to various challenging robotics problems. These problems include ensuring collision-free motion planning in a changing environment, identifying objects that can be picked from the surrounding environment, finding grasps that are feasible for the robot gripper’s size and shape, replanning to a new grasp if the object moves, and returning to a safe location if pick attempts fail multiple times. The project required the Kinova Gen3 Lite robot to run using images from a camera near the robot. However, the off-the-shelf software available for this robot did not support advanced motion planning, making the task more challenging.

Solution

PickNik Robotics specializes in solving challenging robotics problems, making them the perfect choice for this project. They used their expertise in robotics to develop a solution using MoveIt Pro, their C++ plugin-based system, to make the robot system work sustainably and consistently. MoveIt Pro’s robot-agnostic feature made it easy to use existing pick-and-place behavior on a new type of robot. PickNik also created new behaviors such as grasp sampling and cuboid detection, which were not included in the core features. The plugin-based system made it easy to add these new features. MoveIt Pro behavior trees allowed for the re-use and combination of existing robot capabilities to build unique and complex tasks, making it easier to add new features in the future. The resulting solution met the client’s requirements for an interactive robot autonomy demo that could be operated easily by AWS Skills Center staff, showcasing the potential for robotics in the future workplace.

Outcome

The implementation of the PickNik Robotics solution for the interactive robot autonomy demo has been highly successful. The robot arm picks and places objects in the AWS Skills Center. Visitors have been amazed by the robot’s performance, and the project has generated significant interest on social media. The MoveIt Pro-based system requires less involvement from AWS Skills Center staff compared to previous robotics solutions, demonstrating the effectiveness of the solution in creating a semi-autonomous system. The demo has engaged learners, demonstrating the potential for robotics in various industries and opening up possibilities for future applications. Overall, the project has been highly successful, meeting the client’s requirements and exceeding expectations.